Instructions ypp -r

[amolina]

Dear all,

I am doing the steps to speed up the calculations of the BSE ( http://www.yambo-code.org/input_file/ypp/ypp_rim.php ) but I don't undertand the point 7.
Do we have to re-run again yambo with the same input file? I get the same results if I do so.
How can I tell YAMBO to do the calculation using E-RIM?

Thanks in advance,

Alejandro.

[claudio]

Dear Alejandero

>I am doing the steps to speed up the calculations of the BSE ( http://www.yambo-code.org/input_file/ypp/ypp_rim.php ) but I don't undertand the point 7.
>Do we have to re-run again yambo with the same input file? I get the same results if I do so.

yes, but you should get a different result.
Did you have the file ndb.E_RIM in your SAVE folder?

Cla

[amolina]

Ciao Claudio,

OK, I could do for RPA, (options yambo -o c). But not for the BSE calculations (yambo -b -o b ). One has to do something different?

Cheers,

Alejandro.

[claudio]

Ciaio Alejandro

for BSE you have to do

yambo -b -o b -y i

Cla

[amolina]

Ciao,

OK, now it works. Sometimes I got this message:
[BSE INV] Non converged frequencies : 281

Is there anyway to remediate this?

Cheers,

Alejandro.

[amolina]

Hi,

I have been trying different things for this problem with yambo -i

If somebody has had similar problems:

A first option is to reduce the number of BSE bands.

Cheers,

Alejandro.

[claudio]

Dear Alejandro

your output should be something like:

<06s> [06.01] Inversion solver
<06s> Green`s function |####################| [100%] --(E) --(X)
<06s> Freq loop [pert] |####################| [100%] --(E) --(X)
<06s> [BSE INV] Non converged frequencies : 77
<07s> Freq loop [full] |####################| [100%] --(E) --(X)
<07s> [BSE INV] Converged frequencies (after full inv) [o/o]: 100.0000
<07s> [07] Game Over & Game summary


if YAMBO is not able to converge some frequencies with the perturbative approach,
than it tries with the full inversion [full] and if at the end of the run you get the message

<07s> [BSE INV] Converged frequencies (after full inv) [o/o]: 100.0000

the calculation is correct.
Let me know if this is the case

Ciao
Claudio

[amolina]

Hi Claudio,
I got exactly this message when it works. But sometimes stop after the message "Non converged frequencies: 40". Maybe due to memory requirements...

Anyway, the results for Bethe-Salpeter are awesome!

Thanks,
Alejandro.

[claudio]

Ciao Alejandro

infact the full inversion of the BSE equation requires a lot of memory,
I advise you to run in parallel the BSE. You can generate the input for the matrix with

yambo -o b -b

and than solve it

yambo -o b -b -y i

and to run yambo in serial do: yambo -N

you can find here the reference to the method:
http://prb.aps.org/abstract/PRB/v86/i12/e125203

ciao
cla